Fascinating Mindscape this week as Erik Winfree talk about information and dna and robots.
He describes how if you put molecules into a solution and let them settle, they come to equilibrium. But if you heat, cool, re-heat, in a cycle, the molecules can capture energy from the heat into a meta-stable state which, when they are cooling, they can release in a very programmed way.
we have these little dna-units that start assembling into a two-dimensional crystal, during that crystal-growth ... right next to the melting temperature of the crystal which is where you get the lowest error rates, at this point molecules are assembling/dissembling, coming on, coming off, but they're all doing a computation, they're creating that tapestry that is an algorithmic pattern tapestry.
But they are doing it for as close to zero energy as you want, the closer you run to the melting temperature of the crystal, the less free energy difference there is when a piece comes in, vs when a piece falls out, and so the system is naturally exploring this regime Charles Bennett was talking about...
To get the lowest error rates you introduce a form of proof-reading... not which burns energy to do so but a form of self-assemby proof-reading which is based on cooperative interactions. So the growth-process can't go forward unless two things come together at the same time, and of one of them missassembles then the other one won't fit and self-assembly stalls out.
You can prove some interesting things about the kenetic and error-rate trade-off based on the co-operativity (the type of self-assembled pattern)
https://www.youtube.com/watch?v=3ggjfwIty9g